Knowledge Representation

Finke's Principles of Mental Imagery

1. Mental transformations and movements similar to actions on physical objects and precepts 2. spatial relations are similar too 3. can use imagery to infer information that is not semantically encoded 4. mental and perceptual images are constructed 5. functional equivalence

Image scanning

Build up a mental map, move between landmarks using imagery-Reaction time linearly increased with physical distance between landmarks, thus the map is encoded as an image and scanned

Neuropsychological Evidence

Congenitally Blind Subjects are able to perform mental imagery tasks-seems to imply separation of imagery and perception. However, more recent evidence has show that congenitally blind subjects use their visual cortex for spatial representations, Braille reading

Neural Substrates for Imagery and Perception

Cortically blind subjects are unable to use mental imagery-but lesion data is not precise because PET, fMRI, ERP & EEG studies all show occipital lobe activation during visual imagery. Module-specific deficits for perception and imagery

Module-Specific Deficits for Perception and Imagery

Cortically color blind: Ventral Posterior Temporal Cortex LEsions, specific loss of color imagery Double dissociation-localization impairment, parieto-occipital lesion, can draw objects from imagery, but not solve spatial problems Object Recognition Impairment (Agnosia): temporo-occipital lesion, can use imagery for spatial problems (mental rotation), cannot draw objects using imagery (can draw them from perception) Prosopagnosiacs are unable to perform face imagery Visual Spatial neglect patients are unable to visualize scenes on neglect side, but able to on normal side

Functional Equivalence Hypothesis

Imagery is not identical to perception, but it is functionally equivalent. same functions may be performed on imagery as on real percepts. same representational structures. EXCEPT: imagery tends to be lazy, failing to fill in details, and not everyone is good at imagery

3 types of representations in cognitive maps

Landmark Knowledge (points)-information about particular features at a location (BU bridge, the citgo sign) Route-Road Knowledge (lines) directions-specific pathways for navigation Survey Knowledge-global survey map knowledge, with distances

Mental Models

Mental models differ from images in that they permit generalization and reasoning. They are not limited to a specific perceptual-type representation. Mental models differ from propositions because they aren't strictly logical, they are more like a physical model, can have implicit application, and offer greater flexibility than rigid propositions. Mental models are actively constructed and revised-new information is quickly integrated

Mental Rotation

Reaction time increases linearly with angle of rotation for the same pairs. object were mentally rotate in 3-D with constant speed

Heuristics

Symbolic/rule-based/propositional information Right angle-when i turn a corner i think that i turn on my map by 90 degrees Rotation-we tend to distort representations so that directions lie more along NESW axes Relative Position-propositional form of map information-heuristics are propositional-based reasoning approaches

Conceptual-Propositional Hypothesis

Underlying propositions only, abstract, logic-like, images are epiphenomenal, secondary to cognitive processes

Mental Imagery

a mental representation of sensory/perceptual and motor aspects of things which are not currently being sensed; not just visual; it allows us to recall information, solve problems, and plan actions does the mnemonic image actually involve the visual system? YES

Propositional Knowledge

can be expressed as a statement; it is believed to be true and can be used in logical reasoning. High level abstraction-words, pictures represented as deeper meanings, not simply words or pictures; for recall we translate propositions back into words and pictures advantages: very compact code: data compression and good representation for logical reasoning

Dual-Code Hypothesis

form of knowledge representation in images, symbols, either form can be evoked independently

mental imagery summary

imagery provides a useful form of knowledge representation, different than semantic representations. Imagery uses similar representations to those of perception. Mental imagery appears to be modality specific, consisting of several cooperating modules

how is knowledge represented

images (any perceptual modality, largely uses perceptual brain structures) propositions (abstract, logic-like) mental models (intermediate state) cognitve maps largely mental model but can encompass all 2

cognitive maps overall

largely a mental model-formed to explain our experiences, however image-based information may contribute to landmark knowledge, propositional knowledge may contribute to survey based knowledge

Johnson-Laird's Synthesis

mental representations may take 3 forms: 1. PROPOSITIONS-fully abstracted representations that are verbally accessible 2. MENTAL MODELS-knowledge structures constructed to understand and explain experiences 3. IMAGES-specific representations which retain many perceptual features

Acquisition of Cognitive maps

on the knowledge-london taxi drivers train for 2-23 years-intensive study, their posterior hippocampus grows across their training

cognitive map

stored spatial knowledge it is a type of mental model, so giving directions are hard because it is not propositional. images contribute but are insufficient-images are associated with a location. they are internal representations of the physical environment centering on spatial relationships